Every major U.S. health insurer uses artificial intelligence and algorithmic decision support in prior authorization, claims adjudication, and coverage determination. That is no longer in dispute. What remains in dispute — at the regulatory level, at the member level, at the investor level, and increasingly in federal court — is how much of that operational reality each insurer has actually disclosed.
Class action litigation has reshaped the disclosure landscape over the trailing 30 months. The PXDX algorithm class action against Cigna and the nH Predict cases against UnitedHealth Group forced both companies into a level of AI disclosure neither had previously adopted. CMS-0057-F took full effect in 2026. California SB 1120 and AB 3030 require new categories of disclosure for AI in utilization review. The Senate Permanent Subcommittee on Investigations issued its October 2024 report on Medicare Advantage prior authorization denials. The events of December 2024 brought unprecedented public attention to insurance decision processes.
This audit is not a measurement of how sophisticated each insurer's AI deployment is. It is a measurement of how transparently each insurer has chosen to disclose that deployment to investors, to regulators, and to the members whose coverage decisions are being made.
Methodology
Everything-PR audited public disclosure documents from the eight largest U.S. health insurers by enrolled membership: SEC 10-K filings, annual reports, member benefit documents, member portal disclosures, Explanation of Benefits (EOB) language, public regulatory filings, prior authorization criteria documents, and statements to the Senate Permanent Subcommittee on Investigations and CMS during 2024–2026 rulemaking cycles.
Each insurer was scored on six dimensions:
10-K and Annual Report AI Disclosure. Specificity of AI/algorithmic decision-making references in investor-facing disclosure documents.
Member-Facing AI Disclosure. Plan documents, member portals, and explanation-of-benefits materials.
Prior Authorization AI Transparency. Disclosure of algorithmic decision support in prior authorization workflows.
Claims AI Disclosure. Disclosure of algorithmic decision support in claims adjudication.
Algorithm Audit and Third-Party Validation. Public reference to independent algorithm validation, bias audits, or external review.
Member Appeal Process AI Disclosure. Whether member appeal materials disclose that AI was used in the underlying coverage decision.
The composite is the AI Disclosure Score. Maximum: 100. Higher scores reflect more transparent public disclosure — not more sophisticated AI use. The two are distinct measurements.
The Eight Largest U.S. Health Insurers — Ranked by Disclosure
1 Kaiser Permanente — Grade: B | 76 / 100
The most publicly disclosed AI program in the index. Kaiser's integrated payer-provider structure creates a different disclosure environment compared with peer insurers because clinical AI tools appear across member-facing education resources, published Permanente Medicine clinical materials, and research initiatives. The Garfield Innovation Center and the Augmented Intelligence in Medicine and Healthcare initiative are publicly described in significant detail. Because clinical decisions occur inside an integrated structure rather than across fragmented utilization review systems, Kaiser avoids some of the disclosure pressure faced by payer-only models. The principal gap remains at the member-portal level, where AI disclosures remain less detailed than the clinical documentation. The strategic implication is that Kaiser establishes the disclosure benchmark for integrated healthcare systems.
2 Health Care Service Corporation (HCSC) — Grade: B− | 68 / 100
The mutual ownership model has historically produced conservative public disclosure, although measurable improvements have emerged between 2024 and 2026. HCSC's regulatory filings increasingly reference AI and algorithmic decision-making with greater specificity than earlier periods. The acquisition of Cigna's Medicare Advantage business also expanded disclosure responsibilities and increased visibility around utilization-review algorithms. A remaining limitation is that claims-adjudication AI disclosures continue to rely on generalized language. The opportunity remains substantial for a company that historically operated with lower public visibility than larger national peers.
3 Elevance Health — Grade: C+ | 62 / 100
Disclosure improved after the Anthem-to-Elevance rebrand. Recent filings increasingly reference AI and algorithmic support systems while highlighting the Carelon division as the primary AI center for the organization. Although centralizing AI narratives under Carelon creates stronger disclosure consistency, it also limits transparency around how AI functions across the broader insurance operation. Investor-facing communication improved more rapidly than member-facing transparency. A continuing challenge remains the use of broad terminology such as "clinical criteria" without explicit explanation of algorithmic involvement.
4 Humana — Grade: C+ | 58 / 100
Humana's concentration in Medicare Advantage increased pressure for greater AI disclosure because of regulatory developments and broader investigations into utilization review practices. Public filings increasingly describe AI use in utilization management processes with more detail than many peers. The company's movement toward primary-care ownership through CenterWell may also reduce future disclosure pressure by limiting dependence on external authorization systems. However, member-facing communication still relies on broad and abstract language.
5 UnitedHealth Group — Grade: C | 54 / 100
Public disclosure expanded largely through litigation pressure. Legal challenges involving algorithmic decision systems pushed UnitedHealth into a level of public discussion that likely would not otherwise have occurred. Recent disclosures reference algorithmic decision support and associated legal issues, while the broader AI infrastructure inside Optum remains comparatively under-described relative to its scale. The disparity between operational deployment and public disclosure remains significant, and regulatory attention will likely continue increasing.
6 CVS Health (Aetna) — Grade: C− | 48 / 100
AI discussions appear regularly in investor presentations and earnings materials but remain notably absent from many member-facing documents. Aetna's operations use algorithmic support throughout claims and authorization systems, yet public member materials frequently describe these processes using terminology developed before AI became central to healthcare workflows. Increasing state-level legislation and broader federal scrutiny create growing pressure for stronger transparency.
7 The Cigna Group — Grade: D+ | 44 / 100
Recovery from prior litigation remains incomplete. Public legal challenges involving claims-processing algorithms became among the most influential AI-related cases affecting healthcare insurance during the analysis period. Subsequent disclosures included greater acknowledgment of algorithmic decision systems, although ongoing legal considerations continue limiting the specificity of public communication. A central question moving forward is how Cigna chooses to rebuild trust and whether future disclosure becomes proactive rather than reactive.
8 Centene Corporation — Grade: D | 38 / 100
Centene demonstrated the lowest level of public AI disclosure among the largest U.S. health insurers analyzed. Historically, its Medicaid-heavy structure generated lower disclosure pressure than commercial or Medicare-focused organizations. However, evolving state regulations increasingly require greater transparency. Current references to AI and machine learning remain broad and do not provide detailed descriptions of operational deployment. The principal risk is that future regulatory or legal events could force rapid disclosure adjustments.
Key Finding: Disclosure increasingly follows litigation rather than regulation. The next stage of competitive advantage may belong to insurers that choose transparent disclosure before external pressure forces it.
What the Data Shows
Pattern 01 — Disclosure Tracks Litigation, Not Regulation
One of the clearest findings across the analysis is that disclosure improvements have followed legal pressure more consistently than regulatory action. The insurers showing the largest improvement in public AI transparency over the previous thirty months — particularly UnitedHealth and Cigna — expanded disclosure primarily after litigation activity rather than in response to formal regulatory requirements.
Regulatory developments including CMS-0057-F, California SB 1120, Colorado SB 21-169, and Senate investigations all contributed to expanding the broader discussion around disclosure standards. However, the practical mechanism forcing organizational change was litigation itself. Legal exposure created direct incentives for companies to explain operational practices and document algorithmic decision systems more explicitly.
The strategic implication is straightforward: organizations that improve transparency before litigation emerges are likely to incur lower operational, legal, and reputational costs than organizations compelled to change disclosure practices under active legal pressure.
Pattern 02 — The Investor-Facing Versus Member-Facing Disclosure Gap Has Become a Strategic Vulnerability
Across the industry, insurers consistently disclose more information regarding AI usage to investors than to members. Public filings, earnings discussions, and investor presentations increasingly reference AI systems and algorithmic tools in measurable detail, while member-facing materials frequently continue using older terminology that predates current AI-driven operational practices.
The most visible version of this imbalance appears within organizations where investor discussions routinely address AI capabilities while member materials contain little or no direct reference to algorithmic systems. This creates a structural disconnect: companies may satisfy disclosure standards for investors while leaving policyholders and members with limited visibility into how technology influences care decisions, claims processes, or authorization workflows.
State legislation is increasingly targeting this issue through requirements for more explicit member-facing communication. Organizations that proactively improve member documentation before legal requirements force changes are likely to experience lower regulatory friction and smoother adaptation over future regulatory cycles.
Pattern 03 — Integrated Payer-Provider Models Disclose More Because Their Structure Supports It
Integrated healthcare models naturally create broader disclosure surfaces because AI systems become visible throughout care delivery itself.
Organizations such as Kaiser Permanente illustrate this structural advantage. Patient education resources, clinical publications, research initiatives, and care-delivery environments all create opportunities for explaining how AI tools function operationally. These multiple touchpoints produce transparency that fragmented payer-only systems do not naturally generate.
The challenge for non-integrated insurers becomes building equivalent transparency infrastructure without having integrated care environments. Emerging models such as Humana's expansion through CenterWell and broader care-delivery investments across the industry may begin creating new channels for disclosure that historically did not exist.
The strategic question is not simply whether organizations disclose more information, but how they create environments where meaningful disclosure becomes part of normal operations rather than a regulatory obligation.
Pattern 04 — Appeal-Process Disclosure Represents the Largest Industry Gap
Among all evaluated dimensions, disclosure surrounding member appeals and denial processes showed the weakest performance throughout the industry.
When members receive denied coverage decisions and begin formal appeals, documentation frequently provides little or no explanation regarding whether AI systems or algorithmic support influenced the original decision process. This creates a transparency gap at one of the most sensitive points in the member experience.
Because appeal processes directly affect questions of fairness, accountability, and trust, regulators increasingly view this area as a major point of concern. Existing and proposed state legislation has begun focusing specifically on disclosure requirements related to decision systems and appeals.
The opportunity for industry leadership remains significant. The first organization to establish a comprehensive and transparent standard for explaining AI involvement in appeals could define the benchmark for the broader market.
Pattern 05 — Ownership Structure Creates Strategic Flexibility
Ownership models increasingly appear to influence disclosure behavior.
Organizations operating under mutual structures or integrated not-for-profit arrangements experience different pressures than publicly traded insurers. Because these organizations face reduced quarterly investor expectations, they often have greater flexibility in determining how aggressively they pursue transparency initiatives.
The highest-performing organizations in the analysis also reflected less conventional ownership structures. This suggests that governance design itself may increasingly influence disclosure leadership.
The broader implication is that ownership structure is no longer merely a financial characteristic. It increasingly functions as a strategic variable influencing how organizations approach transparency, regulatory adaptation, and long-term trust building.
What this means
The health insurance industry is in the early phase of a multi-year transition in which AI and algorithmic decision support — already deployed across prior authorization, claims adjudication, coverage determination, fraud detection, and care management — will be required to be disclosed in member-facing, regulator-facing, and investor-facing documents at substantially greater specificity than current disclosure practice. The transition will be driven by three forces: state-level legislation (already enacted in California and Colorado, pending in multiple other states), federal regulatory action (CMS-0057-F implementation, Senate Finance Committee inquiries, potential CMS rulemaking on AI in Medicare Advantage), and continued class action litigation pressure.
The insurers that have led the disclosure transition have done so in response to specific external pressure — litigation or regulation. The insurers that lead the next phase will have to do so proactively, before the external pressure compels them. The strategic case for proactive disclosure is operational, not philosophical: proactive disclosure produces a controllable comms environment, while reactive disclosure produces a crisis comms environment. The cost differential between the two is measured in years of recovery rather than quarters.
The forward-looking question for 2026: which of the eight largest U.S. health insurers will be the first to formalize a comprehensive, member-facing, third-party-audited AI disclosure standard — and accept the short-term comms cost of being first in order to capture the long-term comms benefit of being the industry benchmark? The data in this audit suggests the structural advantage sits with the integrated and member-owned models. Whether the investor-owned national insurers will move before they are compelled is the strategic question of the next 24 months.
Submissions and Methodology Inquiries
Submissions, methodology questions, and comment from insurers:
